Joe Germuska. There you go, Joe. Thank you. We're going to be three, and you'll see about that in a minute. But hi everybody, I'm really happy to be here with you. My name is Joe Germuska. I'm the chief nerd at Knight Lab, which is the media technology and design studio at Northwestern University, just outside of Chicago in the United States. Knight Lab is a we build tools for journalists and storytellers, we do design research, we promote product thinking in journalism, and of course we teach students. My professional background is actually more in internet software development than journalism, but I got into this field at the Chicago Tribune.
I did a stint as a news applications developer there, and then I came to Northwestern uh to join the lab. And I'm super excited to be here. I was super excited to see a whole track about journalism. When we pitched this talk, I was like, okay, well, we're gonna explain this to people. And now there's people all day have already been saying smarter things that I'm gonna tell you today, but anyway. And uh oh, by the way, I realized that I have had a cute or whatever uh avatar, and I've been noticing recognizing people at the conference from their so I thought I should throw mine on there because it's not me, so that's me if you if it is familiar.
Anyway, also uh I'm co-presenting with Ben Werdmuller. Thank you so much, Joe. I'm so sorry I can't be there in person. It's for reasons right outside of my control, which is why I'm a disembodied voice right now, coming to you like a Timu Map Barry. My name's Ben Wodmiller, I lead technology at ProPublica, a nonprofit newsroom in the US that investigates abuses of trust and power in the public interest. I've also got a long history of building on and writing about the open social web, and I'm really excited to almost be here with you all right now.
We'll hear more from Ben later. So but my I have the first part. So I mentioned one of the things we do at Knight Lab is design research, and over the last few years we've been involved in a series of projects that we call next gen news. Uh we were partnered with the Financial Times Strategies and were supported, I need to say, by the Google News Initiative. Thank them for this. Um the idea was to understand the news and information behavior habits of younger adults, uh, 18 to 25 especially, seeing them as kind of the vanguard of where things are going and hoping that if we can characterize how they interact with information these days, maybe publishers can start to think better about how to meet them in the next few years and sort of rearrange their process to be better serving the needs of that audience.
So in 2023, we did about four dozen one-on-one interviews in the United States, Nigeria, and India. And then last year we did another round of qualitative research. We did about 85 diary studies in those three countries, plus we added Brazil and the United Kingdom. We also did a quantitative survey, uh, 1,000 people in each of those countries, uh, and the quantitative survey went from 18 to actually 101 years old, so we could do some analysis that sort of compared across brackets and buckets. Um so um just some of the top line findings from that that frame how we're thinking about this stuff and maybe might set up uh thinking about how news has to change.
These are things that this audience is probably familiar with, but not all publishers are, and still just also being able to validate it. Say we actually went out and found um one of the huge things with this audience is a continued sense of content overwhelm. Um the world has gone from finite content where you could start at the front page and know when you flip the last page that you were done, and now we have an infinite stream. Um we went from a situation where news was defined pretty narrowly and basically defined by the publishers to a world now where news could be anything.
There's so many things that are volume for people's attention, and so uh publishers have to recognize the very different landscape that they're moving into. Another thing that we think is really interesting is the way that sort of quality markers and trust, which we'll get to on the next slide, uh have kind of fallen away because now everything kind of looks the same. It's all on a screen, it's all in basically the same shape, and so it's a lot harder now for people to just know what exactly it is that they're getting when they are consuming information.
And so that also leads to a very strong, you know, over and over from these dozens of young people we've talked to. Trust is understandably a big issue for them. They're not been served well by a lot of media institutions. Um they're concerned that news is gonna be false, that it might be sensationalized, that it's biased, it might be combinations of those. And so what they really want is a sense from the sources for their news of credibility, affinity, and transparent intention. Um just like a factoid that maybe we can sort of verify, it probably fits everyone's assumptions.
But we find that um almost 76% of the youngest bracket of the audience get their news from social media sometimes or often. So this is a question where people could answer as many as they wanted to. Um but more than 50% of people up to 50 get uh often get news from social media. So obviously, this is a huge change in consumption and a big place to be conscious about where we're meeting audiences. Another thing that this audience uh maybe less when we realize how many people who are familiar with journalism are here, but still a lot of people from the AT Protocol world and the engineering world may just overestimate what newsrooms are working with or underestimate the scope of the problems.
You know, the bottom fell out of advertising in journalism uh really 20 years ago uh and it hasn't gotten any better. Um consolidation layoffs, the threat of lawsuits are hitting all over the place. Google search for filler referrals to publishers dropped 33% globally in a single year from 2024 to 2025. AI overviews are at the top of almost 10% of US search results. Um when they those show up, only about 1% of users click through. Facebook referrals to news sites are down like 80% since 2020. Ex referrals have dropped about 60%. Um small publishers have been hit the hardest.
Social referrals for smaller outlets dropped 98% over five years from 10.1 million page views to under 187,000. And in just four years, this is uh from the Reuters Institute uh Reuters does a lot of great research into the state of journalism from their uh journalism technology trends of this year. They found just in four years, uh there's a 22 percentage point drop in people who were confident about the prospects for news. Um people are starting to realize that it is dire out there. Obviously, economic sustainability is a key issue. As I said, advertising is bottomed out, really only works at very large scale.
Platforms take a huge cut, uh, and of course the algorithmic wins shift, you're in a lot of trouble. Membership works for a business model for a lot of organizations. Uh The Guardian is one that's doing fairly well with it. Um but it does really require leaning into relationships, developing that developing that affinity that we talked about earlier, um, cutting through the content overwhelm and making sure that people value you enough to go to the trouble to join. Um there's promising science, but it's a big question, especially about how large organizations that were built around a different economic model are gonna adapt to this world.
And then email newsletters were maybe going to be the last way that you could really own your relationship with audience. But as AI starts creeping into people's inboxes, the likelihood that people are reading at depth uh is even uh uh less than ever, so that channel is also at risk. So newsrooms most likely to lose their audiences are the ones covering the most underserved communities. So Ben lives in Philadelphia, as he observed the inquirer could run citywide ad campaigns and sort of keep their um face in, you know, keep themselves in people's attention. But smaller places like the Kensington Voice, sort of an underserved part of Philadelphia can't do that.
So independent voices, while there's a lot more out there really have a lot of platform dependency challenges and a lot of just sort of cutting through the noise to get people's attention. Centralized platforms or silos are unreliable partners. AT Protocols for Publishers in London this February, Andrew Winnensland documented what she called a brief history of algorithmic fuckery, a timeline of every time platforms pulled the rug out from under the communities they were supposedly supporting. Here's an example. In 2015, Facebook launched instant articles. The New York Times, the Washington Post, BuzzFeed, and dozens of others invested heavily.
In a 2018, Facebook announced a major algorithm change, deprioritizing news. And by 2023, instant articles were shut down entirely. Publishers who restructured their operations around the product were left with nothing. Here's another infamously. If you mention the phrase pivot to video in any newsroom, people will physically shut up. All of this has happened before, and all of this will happen again. This repeats because the incentives are baked in. The platform's customer is the advertiser. Publisher content is the bait. When the platform no longer needs that bait, the publisher is stranded. Well, uh sorry for the jump scare, I guess.
So right now, platform algorithms are black boxes that prioritize business goals over audiences or society. A small number of powerful people like this guy with vested interests own the major platforms. Journalists are being doxxed. News organizations are leaving. NPR left X because of bad treatment, for example. In the worst cases, these platforms are culpable for genocides and the undermining of democracy itself. And these platforms don't just control the reach, they control the money too. They can change the terms whenever it suits them. It's all in their control. But as everyone in this room knows, all is not lost.
While links are suppressed on many platforms and sharing them is impossible on others, forcing newsrooms to optimize for vanity metrics that don't help their businesses or really inform anyone, many newsrooms have found that atproto users are disproportionately engaged. Referral traffic, click-through rates, donations from Bluesky are all much higher than from legacy platforms. This audience already acts differently. Feeds on atproto are open infrastructure. Anyone can build an algorithmic feed and users choose which to subscribe to. People have built custom feeds that surface journalism and ensure real reporting reaches real readers. Publishers don't have to be passive consumers, subject to platform decisions.
They don't have to let technology happen to them like an asteroid. They can build. And of course, the open social web is bigger than one protocol. Flipboard has federated via ActivityPub since late 2023, with over a hundred publishers like Axios, The Verge, Semaphor, Smithsonian, Fast Company, the 19th, federating their content through it. A commercial company looked at the trajectory of centralized social media and bet on open federation. It's your identity, your data, your relationships belong to you, not a platform. Nobody can pull the rug. This is the future of media. Thanks, Ben. So of course there's so much more we can build.
And this is really what this talk is about, and that's why we're actually not going to be, I'm not going to be talking that much more. We really want to have a conversation. Our whole goal with coming to Atmosphere Conf was to say there are some people in the news industry who are recognizing that there is a lot that needs to change, and we want to think about it, talk about it, and work with this community to build. I'm especially lucky working at Northwestern that I can just sort of say I should be spending my time on this initiative, and if there's things that we find that we can cooperate on to help make happen, that would be great.
So we came up with a few sort of design provocations, some things that maybe especially people who are more on the engineering side and less familiar with journalism. Maybe this will plant some seeds, help you think a little bit about the problems that journalism has that have not been in the front of your mind. And once we talk through a few of these, we can have questions now, we can talk more later because what we really want to do is solve problems. So one idea is what if you could have like a membership card on your PDS?
You know, it's uh old tradition that people who are like members of public media, at least uh you know in Chicago, for example, you get your WBEZ card, and then maybe you get a discount at a restaurant or something like that. So, what if we could take advantage of the sort of cryptographic trustability that um a person is a member and the the news organization can put a thing on their PDS that says so or otherwise wire things up? Maybe it would unlock different paywall treatment. Maybe this would be a way to bundle subscriptions. A lot of people talk about the risk of subscription fatigue and people just not being comfortable making so many different financial arrangements.
So maybe something like this card could give you privileged access to a partner for a period of time. Um, your membership travels with you across the protocol instead of being locked up in an app. Another huge question: how can we build more censorship-resistant journalism? Uh, you know, in a world where social is powered by open protocols, platforms can't exactly suppress a newsroom's reporting, and they can't exactly turn off the algorithm to fully bury a story just because it's inconvenient. But governments can attack a PDS provider or an individual application. Um we need platforms to be resilient.
So this matters right now, of course, and it's only going to matter more in the near future. Where are some things that we might do to help improve the resilience against censorship for for media? Now for Ben. And what if publishers and their allies could create and share algorithmic feeds designed to service good journalism? Andre Winnensland's news specific feeds on Bluesky are a fantastic proof of concept. And the tools already exist. Grays lets anyone build and publish custom Bluesky feeds. Flipboard Surf App also lets you build custom feeds, and it's a social browser that pulls together Bluesky, Mastodon, and RSS sources into curated feeds organized by content type.
These tools are fantastic. Imagine newsrooms using them to release their own vetted algorithmic feeds built from sources and topics they trust, operating over long-form content as well as short posts. Readers subscribe to the feed, and the newsroom's editorial judgment becomes the discovery layer. And then projects like standard.site are doing fantastic work getting full article content into the protocol layer so multiple apps and algorithms can surface it properly. But how do you let an algorithm read the article so that it can recommend it while still preserving the publisher's right to get paid for their work. This is a problem we can work on together.
And then let's talk about microcommunities inside. I can read slide notes. The Bristol Cable, a reader-owned local newspaper in Bristol built a new app with the Newsmaster Foundation that combines news delivery with a full Mastodon powered community layer. Members get integrated stories and podcasts alongside a single news amps, news platforms, and news websites. The Bristol Cable, a reader-owned local newspaper in Bristol, England, built a new amp with a Newsmast Foundation that combines news delivery with a full Mastodon powered community layer. Members get integrated stories and podcasts alongside a social following section where they connect with cable staff and each other, plus curated topic-based channels.
Because it's built on the Fediverse, the community is interoperable. Members from other newsrooms or anyone on the Fediverse can participate through their own apps. A Bristol Cable member and a member of another local newsroom can be in the same conversation. The community builds trust and the cable aims to double its membership by 2030. The community app is absolutely central to doing that. I forgot that I meant to go through and cache all the audio before I was a surprise. Sorry about that. Anyway. So the gap between what open protocols can do in theory and what a non-technical publisher can use today is really the problem to solve.
And so if you're here for Glenn's talk just before, there's interesting things happening in the Oakland area about ways to apply these things. If you stick around for Android's talk, there'll be even more about things we can do. But it's still too hard right now for newsroom to set up and run community on the open social web. Community management at scale needs moderation tooling. That really doesn't exist yet. There's promising signs, but there's work to do. Publishers need analytics for community health. This is really important when you're trying to make a bit going business concern, even if you're just trying to be sustainable, not rich.
You need to understand how your community's health works. And those metrics are different from traditional web analytics, which are also really important for these audiences. So again, our ask here is for the atproto development community to work with publishers, not just building for them. This is real, there's real mutual interest between the publishing and protocol communities and also gaps in the understanding. Publishers often don't know what's technically possible. Protocol builders don't always know what publishers really need. So sustained collaboration is really where both sides shape the project to get project together. That's going to be how this works.
So let's work together. So that's it for the slides. Now I would just love to hear from folks either ideas you have, questions you have about how journalism works if you're less of an insider. What can we make together. Excellent. Thank you, Joe, and thank you, Ben. Wherever you may be.
So I saw some hands. Excellent. Let's open up the floor for questions. All right, we got one right back there. Um hi, thank you. Um you mentioned uh community primitives in like a couple slides back. I'm wondering if uh newsrooms have any specific vision of what an incredible community around their work looks like. Um yeah. Hi, I'm Celine. I work um on Leaflet. Oh yeah. Hi, Celine. Um I doubt that there are a lot of newsrooms who have done that. I don't recall being in conversation with them about it. There's definitely been um I think more than just anecdotal evidence about the value of, for example, uh journalists showing up in comment threads and participating there.
You know, comments, the way they have traditionally worked on, especially directly on news sites, tend to be pretty bad. Um, but there's a lot of value to them showing up and engaging, it helps to add credibility and uh and again we're talking about the sort of sense of affinity and belonging that we are we want to be a member because you're our people, we want to give you money. But I'm not sure what those primitives are. Um if anyone else wants to chime in on that we can hear, or just sort of noting that as a uh topic for future exploration.
Um I think one of the things that I think is, and I've heard this you know in other talks here, uh, it's a balancing act because I don't think that most of the challenges for journalism are technical. I actually I came into this field at a time, this is you know in the early teens, where everyone was like, oh, the internet has destroyed the news industry, so let's get the computer people in here to fix things. I mean so my backgrounds and software, you know. And um, I mean I'm glad that I got in. A lot of really cool people joined the community, but also it didn't take that long to realize that that wasn't really the place to put.
And that's why a lot of times I sort of I don't know if I say it, but I really think about the problem more as a design problem than a technology problem. But it is awesome that we have these technology building blocks coming up. The short answer is that's a good one. I don't know. Okay.
Thanks. Thanks for the talk, Joe. I missed the beginning. So apologies if I asked you something you ready to touch on. How do you deal with the threat of AI? What are some of your thoughts on you know Cloudflare's paper crawl program? I know Microsoft has a similar content marketplace for publishers, would be keen to get your perspectives on those. Yeah, that's a good question. I think you know, there was one piece maybe before you came in where we acknowledge that like uh instant AI on Google search is really uh hindering click-through. This is kind of well understood.
Um there are people, including like friends of mine in the industry are like people are just going to turn to Chat GPT or whatever, and that's where they're gonna get their news, they're not even gonna go to a page. I'll say, and I am known to be a dyed-in-the-wool idealist, but I really think that the human knowing the human factor is one of the secret weapons here. I think that when we one of the one of the earlier slides that was at the beginning is the desire that especially young audiences, but really everyone has trust and a sense of affinity.
And so I think that there's a lot of value to knowing where your news is coming from. And I also think I feel like I haven't heard this that much, and I don't have it perfectly articulated, but I feel like the problem with uh asking an AI for your news is not that different from the problems with letting an algorithmic feed direct your attention. It's like, well, what decisions have they made and how is that interfering with what you're getting? Are you actually getting what you want? And I think uh because of just the mystery of AI, it's gonna take the general public, I think, a little while to really see that as a risk.
But on the whole, I guess I would just say we have to lean into there is a market for this. And there's you know, one of the big challenges with this also uh, you know, I worked at a big news organization. I think our partners in this next-gen news research are mostly thinking about big organizations. That's not personally my stake. I am interested in anything that helps audiences and people who have a thing that they want to say, who want to make their living, being a person who makes information, to have them connect with less interference, which is why I'm so excited about AT Protocol.
Um I think so uh it may not be replacing the biggest feeds. There's gonna be uh or biggest industry or organizations, um, there's always gonna be a mass audience that chooses kind of the easy route and so on, and I don't know that we can do that, but if we can make it sustainable enough for the people who choose a different path to be able to do their thing, I think that's a win. So, in short, I think that there's the market for people instead of AI. And I there's also lots of not just hand waving about how AI might make it more efficient to do the work.
Um, but AI generated content I'm not long on personally. Um you mentioned community engagement tools, and um I'm asking because I used to work on an open source commenting tool called Coral that was adopted by a lot of um newsrooms. Where do you see the incentive for publishers to adapt their existing tools to like the open protocols versus adopting new tools that are getting created? Yeah, I mean it's a good question. Uh I uh Coral has done great research, thank you for whatever part you played in that. I mean, I love I knew a lot of people on that team and uh was proud of it.
The sponsors of it don't really seem to have even uh bought into it that much. So the motivation is it's a good question, and I think this is another case where um there may be a little bit of uh dinosaurs and small mammals thing where we're in a phase change, and some of the big people who don't want to change don't, and they may not make it through. Um I think that especially when you're a large organization, you have to balance um. So, where was this uh it's been a full day people were talking about the Washington Post and uh and how hard it was to get them to like try new things.
Um yeah, it was um Laura and I forgot to mention, so we part of the NextGen news research was interviewing news creators, uh, including Lauren Sachs, who was on stage in an earlier session, uh she works with local news international, and they're kind of the outshoot of Washington Post's briefly quite lauded TikTok presence, and then the post was like we can't pay you all and they took the buyout and they went and did their own thing and they're doing great and they can experiment and they can switch. So if the big guys can't switch I don't know.
I mean there's something to be said for making the case clear you know for them to make changes but especially if you have the the experience you have at Quoral if you were unable to persuade people that this is a better way you should follow it then I'm not going to say that I know a better way than that. So you know there's a gonna be a there's a lot of shaking out happening so you you talked about kind of how we we design for getting readers to to trust the news sources and but it kind of seems like it at this particular moment in time one of the biggest threats is how do you make news organizations resilient to somebody with just a shitload of money coming in and saying I don't want you to do it this way anymore which which seems to be more of what's kind of hollowing out news in general.
You mean uh like Alden publishing buying the Chicago Tribune shortly after well that was a little after I left but yeah um or Jeff Bezos and the Washington Post. I mean I think that there's not much you can do when people with money can buy things. In my uh heart of hearts enough people will know that that is a lesser product. And I mean the Washington Post bled hundreds of thousands of subscribers when they decided not to do an endorsement in 2024. The audience that expected them to be sort of adhere to all these old-fashioned news ideals saw that they were like oh we can't put in the editorial up that is against the guy who's probably going to win and if he wins he's gonna take it out on us if we did um and I mean I canceled my subscription then and I'm again hundreds of thousands of people did.
So I do think that that will out. Again I'm a died in the wool optimist but I think that and I mean a lot of this buying is happening at like the broadcast level I mean CBS news is tanking it's like all over the headlines it does it's not working. Money is powerful they may be able to do more than I wish they could but what I really want to see is the people who have the ethos that I believe in again be able to sustainably do the work I take a lot of 404 media as one of my favorite stories because they just do brilliant work and it's like five people and they are making a go of it and they they had I think their second anniversary recently and they the story they tell is that they're they're doing all right and there's been a lot of other ones like that.
So I think that's more what I'm interested in not only small but basically make it possible to do the work without having to have the money already when you start.